Many Analysts, One Dataset is a crowdsourced metascience project: many independent research teams analyze the same dataset and research question, then compare how their (often equally defensible) analytic choices change the results. I was one of the participating analysts and a co-author on the resulting paper.
Twenty-nine teams asked whether soccer referees are more likely to give red cards to dark-skinned players than light-skinned ones. Estimated effect sizes ranged widely; neither prior beliefs nor expertise neatly explained the variation. The point isn't that any one analysis was wrong—it's that complex data leave room for subjective choices, and those choices matter enough to make transparent.
More about the project: